Functional Magnetic Resonance Imaging

The advent of fMRI in the 1990s allowed clinicians to examine changes in brain activity after a stereotyped task ( ). The first imaging methods used intravenous gadolinium contrast or the BOLD technique, which detects levels of deoxyhemoglobin in stimulated brain ( ). When subjected to a simple mechanical or visual task, areas of eloquent cortex become activated and subsequently utilize oxygen to generate energy. As a result, local levels of deoxyhemoglobin increase in concentration, and these depots of deoxyhemoglobin have inherent paramagnetic properties that can be detected with BOLD imaging ( ). Since then there have been significant advancements in the imaging capabilities of fMRI, including the ability to examine brain activity at rest with rs-fMRI ( ). When compared to BOLD imaging, which depicts brain activity during stereotyped tasks, rs-fMRI demonstrates brain activity without an applied stimulus, or at rest. Although intraoperative, direct electrical stimulation remains the gold standard for mapping functional cortex, these new fMRI technologies allow better preoperative planning and intraoperative guidance.

Fomberstein et al. described several anatomic regions in the brain that have been implicated in pain perception ( ). These areas, termed the “pain matrix,” include the primary and secondary somatosensory cortex, anterior cingulate cortex, and insula ( ). Abnormal functional connectivity between these areas has been shown to be highly correlated with several chronic pain syndromes ( ). The authors found that functional connectivity between the prefrontal cortex, insula, and anterior cingulate cortex could be used as a marker for chronic pain that correlates highly with poor treatment response to opioid therapy ( ) ( Fig. 46.1 ) . The authors noted several interesting patterns of activation, displayed on fMRI in several chronic pain states. For example, patients with chronic low back pain (CLBP) had increased activation of the medial prefrontal and anterior cingulate cortex when compared to normal controls. After 2 weeks of treatment, connectivity of the dorsal median prefrontal cortex and insula was seen in the treatment-refractory patient group ( ). In addition, higher connectivity between the dorsal median prefrontal cortex and insula was directly correlated with decreased responsiveness to placebo treatment in this study ( ). These results support the long-held belief that chronic pain syndromes are a multifaceted entity involving more than just noxious stimulus of peripheral or autonomic nerves ( ).

Brain regions involved in (A) pressure pain processes and (B) pressure pain > nonpainful pressure as measured by BOLD fMRI from simultaneous PET/MRI scans. BOLD , blood oxygenation level dependent; PET , positron emission tomography; MRI , magnetic resonance imaging.

Adapted from Wey, H.Y., Catana, C., Hooker, J.M., et al., November 15, 2014. Simultaneous fMRI-PET of the opioidergic pain system in human brain. Neuroimage 102 (Pt 2), 275–282, with permission.

As most clinicians are well aware, patients with complex chronic pain syndromes can be quite challenging to manage and often require multidisciplinary care. These results suggest that fMRI techniques could be used to predict treatment response to analgesia, thus creating a framework to build patient-specific analgesic regimens that are supported by connectivity. Connectivity could be used to predict opioid response and probability of treatment success/failure, set thresholds for treatment duration, and monitor treatment progression. Unlike conventional MRI, fMRI can be used to obtain objective information about subjective clinical data, such as pain. With this in mind, there have been more recent efforts to tailor fMRI sequences to detecting areas that become active after a noxious stimulus, in hope that this data can be used to “diagnose” pain.

In 2013 Wager et al. were the first to develop an fMRI-based neurologic signature that could detect physiological pain. In response to a noxious stimulus, Wager’s pain signature demonstrated increased activity in the primary and secondary somatosensory cortices that was over 90% sensitive and specific for the experimental noxious stimulus ( ). The study investigators hypothesized that the signature could be applied to clinical situations when patients are unable to communicate pain or clinicians suspect incorrect self-reporting of pain.

In another study conducted by Loggia et al. in 2014, fMRI was used to characterize pain physiology and circuit pathways in 13 patients with fibromyalgia. The investigators found that these patients repeatedly experienced an exaggerated response to a painful stimulus when compared to normal volunteers. In addition, fibromyalgia patients demonstrated reduced reward/punishment signaling, thought to be due to decreased activity in the periaqueductal gray and ventral segmental areas of the midbrain. They hypothesized that suppression of activity in these regions may explain why fibromyalgia patients do not respond well to traditional pain therapies ( ). In fact, the treatment of fibromyalgia often requires a multidisciplinary clinical approach and the judicious use of neuroleptic, antidepressant, and occasionally antiepileptic mediations ( ). The results of these studies emphasize the difficulty in treating chronic pain disorders with opioid medications alone, and may suggest that a multimodality approach is needed.

Spinal Cord Stimulation

Rs-fMRI has also been examined in the context of characterizing pain relief from spinal cord stimulation (SCS). Deogaonkar et al. utilized fMRI in 10 patients who underwent thoracic SCS implantation for complex regional pain syndrome or neuropathic leg pain ( ). Patients underwent imaging in both the on- and off-stimulation states. The investigators found that pain relief from SCS reduced resting-state connectivity in the left frontal insula, right primary/secondary somatosensory cortex, and default mode network (DMN) ( ). The investigators further hypothesized that SCS exerted its analgesic effects through promoting decreased connectivity between the somatosensory and limbic areas of the brain. These results are intriguing, because they suggest that the effects of SCS are more complex than simply blocking the propagation of pain stimuli from peripheral nerves. In fact, the effects of SCS trigger changes in the severity and the brain’s processing of pain at the level of brain circuitry.

Although these studies suggest that fMRI techniques may be utilized to characterize pain objectively, clinical investigators underscore the need for further studies with larger populations before determining the clinical utility of these sequences for pain syndromes. Functional imaging techniques have been used for identification of eloquent areas in tumor and epilepsy neurosurgery, but the application of these techniques in neuromodulation is still under investigation ( ). Certainly, as more advanced and pain-specific fMRI techniques are developed, functional connectivity patterns between various pain-generating areas of the brain could be better detected and treated ( ).

Arterial Spin Labeling

Traditionally used in the imaging of brain tumors, the ASL MRI sequence characterizes changes in cerebral blood flow (CBF). In 2011 Wasan et al. demonstrated measurable increases in regional CBF to the somatosensory, prefrontal, and insular cortices in patients with CLBP subjected to a straight-leg-raising test. The authors also discovered that the same straight-leg maneuver did not create detectible ASL changes in healthy age-matched controls. Patients with CLBP also demonstrated decreased connectivity between the medial prefrontal cortex and the DMN ( ). The authors hypothesized that higher levels of connectivity between the medial prefrontal cortex and the DMN in the control population may enhance pain-coping mechanisms.

In a study exploring the role for pulsed-continuous ASL (PC-ASL) in a thermal stimulation paradigm, Maleki et al. compared BOLD techniques to PC-ASL in 12 healthy subjects ( ). Using PC-ASL, the authors found specific activation patterns in response to a thermal painful stimulus. The insula, thalamus, hippococampus, amygdala, anterior cingulate, primary and secondary somatosensory cortex, precentral gyrus, and precuneus all had recognizable activation patterns ( ). The authors conclude that PC-ASL may be used to study acute pain, and hypothesize that this imaging modality might provide a more suitable imaging approach to measure brain activity in patient populations where blood flow, volume, or oxygen extraction are altered ( ).

More recently, ASL imaging has been utilized in an effort to identify pain-specific regions of cortical activation and apply ASL to clinical scenarios. recognized the dorsal posterior insula as a fundamental structure for human pain processing, and posited that it may be a potential therapeutic target. Using controlled, slowly varying nociceptive input with a capsaicin-induced thermal stimulation model in 17 awake, healthy human subjects, the authors quantified cortical activation using an optimized ASL method. They subsequently evaluated the correlation between absolute CBF and pain scores during the experimental time course, and determined that, from a whole-brain analysis, only the contralateral dorsal posterior insula showed statistically significant changes ( ). The authors further clarify the putative role of this brain region for tracking pain intensity, and contribute to an evolving discussion about the role of ASL in elucidating the precise circuitry of pain ( ).

Combining ASL data for detection of pain-related changes in CBF with multivariate machine-learning techniques for analysis of prognostic and diagnostic markers of pain, O’Muircheartaigh et al. demonstrated the feasibility of time-efficient probabilistic prediction of clinically relevant pain ( ). The authors assessed preoperative and postoperative ASL data and pain scores in 20 male volunteers requiring bilateral lower-jaw third molar extraction. The data from all sessions was fed into an independent Gaussian process binary classifier that successfully discriminated postsurgical from presurgical states with nearly 95% accuracy. The authors concluded that multivariate pattern classification methods can be applied to ASL data to provide clinically relevant predictions regarding moderate to severe postoperative pain. On the basis of these findings, they further suggest a role for functional neuroimaging as an objective complement to subjective pain measurements in a perioperative patient population ( ).

Magnetic Resonance Tractography

Numerous investigators have observed recurrent patterns of anatomic and physiologic changes in white-matter circuitry of patients with chronic pain ( ). Magnetic resonance diffusion-weighted imaging (dMRI) has been used to explain and further characterize these differential structural patterns. Furthermore, dMRI can be used to estimate the trajectories of white-matter fiber pathways, a technique known as fiber tractography. Experimentally, dMRI has been shown to be a sensitive diagnostic imaging tool to diagnose a wide variety of chronic pain disorders, ranging from lumbar radiculopathy to migraine headaches ( ).

Tractography assesses the restricted random diffusion of water molecules in axons to characterize their microstructure. It is generally composed of the aggregate information of two different imaging sequences, diffusion tensor imaging (DTI) and fractional anisotropy (FA). DTI detects the movement of water through axons and FA characterizes their direction. The structural connectivity of the brain can then be reconstructed and visualized with dMRI. Abnormal patterns of white-matter circuitry, implicated in the development of chronic pain disorders, can then be diagnosed and treated. For example, Kovanlikaya et al. found that pain fibers traversing the nucleus ventrocaudalis of the thalamus could be visualized with dMRI and used as a target for deep brain stimulation (DBS) surgery ( ). Mansour et al. observed differences in the patterns of structural connectivity in the prefrontal cortex of patients who developed CLBP after an inciting event. They hypothesized that these structural differences, characterized with dMRI ( ), which existed prior to the noxious stimulus could predict the patient’s predisposition to chronic pain.

In addition to examining spinothalamocortical pain pathways, dMRI has been shown to be a sensitive indicator of spinal cord and nerve root compression ( Fig. 46.2 ) ( ). Shi et al. found that the mean FA values in compressed nerve roots were significantly decreased when compared to normal uncompressed nerve roots. These findings were further supported by Oikawa et al., who hypothesized that dMRI may offer quantitative values for the severity of nerve root compression, which can be used as a diagnostic tool for lumbar radiculopathy ( ). Similar observations were used by Fujiwara et al. to image and diagnose trigeminal neuralgia from neurovascular compression. In this study, patients with trigeminal neuralgia from suspected neurovascular compression demonstrated lower mean FA values compared to healthy controls ( ).

Fusion image made up of the diffusion tensor tractography of the lumbar nerve (L4, L5, S1) and T2-weighted imaging. Interruption in the right L5 nerve at the foramen ( white arrow ).

Adapted from Eguchi, Y., Ohtori, S., Suzuki, M., et al., February 2016. Diagnosis of lumbar foraminal stenosis using diffusion tensor imaging. Asian Spine J. 10 (1), 164–169, with permission.

This data provides clinicians with an objective, quantitative tool to measure the degree of nerve root compression throughout the lumbar spine. In disease states such as degenerative lumbar spondylosis, where there are numerous and varying severities of nerve root compression, FA values could potentially be used in tandem with clinical exam and history to guide candidate selection for surgery. In addition, FA values can be utilized to quantify and assess the degree of decompression after surgery and potentially be used as an outcome marker for prospective outcome studies.

Chronic Low Back Pain

Over the past decade the structural and functional imaging signatures of brain physiology in patients with CLBP and its modulation and treatment have been a growing area of investigation. CLBP and other types of chronic pain have been associated with both anatomical and functional changes in the cerebral cortex. Several imaging and neurophysiological studies have implicated the dorsolateral prefrontal cortex (DLPFC) as one common site of such changes. For example, using an fMRI-based protocol, Seminowicz and Davis demonstrated in 2006 that pain catastrophization, i.e., the maladaptive exaggeration of a painful experience, was not related to activity in somatosensory cortices of the parietal lobe, but rather to cortical regions such as the dorsolateral prefrontal and insular cortices that subserve affective, attention-related, and motor aspects of pain ( ). Their findings were posited to support the framework of an attention model of pain catastrophization where, despite engagement of a cortical vigilance network during mild pain, diminished prefrontal cortical modulation inhibits pain suppression during more intense painful stimuli. Furthermore, the authors speculate that catastrophization may underlie the mechanistic progression from acute pain to chronic pain.

In 2010 Fierro et al. reported on an exploration of the effect of capsacin-induced pain and the modulatory influences of left DLPFC transcortical magnetic stimulation on motor corticospinal and intracortical excitability; the authors conclude that tonic pain exerts a modulatory influence of motor intracortical excitability and that left DLPFC stimulation can have analgesic effects on otherwise painful stimuli. In the same year Krummenacher et al. reported on the ability of DLPFC stimulation to interrupt placebo analgesia in a heat-pain paradigm ( ).

There is also mounting evidence that some people with chronic pain, such as longstanding back pain, exhibit abnormal cortical function and cognitive impairment. Findings from a 2007 study by Seminowicz and Davis suggest that the demands placed on pain-associated cognitive networks may underlie this cognitive impairment ( ). In 2011 Seminowicz et al. reported on the findings of a prospective study assessing the reversibility of neuroanatomical and functional abnormalities in patients with CLBP, and the dependence of these changes on treatment outcomes ( ). In their longitudinal study, the authors utilized 3 T MRI for structural and functional brain imaging in patients with CLBP. In the experimental group undergoing intervention (n = 14), the imaging studies were performed prior to and 6 months after either spine surgery or interventional pain procedures. The experimental cases were compared to 10 healthy controls. Cortical thickness analysis was performed using structural MRI studies, and participants also underwent fMRI scans during the performance of a cognitively demanding task. The authors found that CLBP was associated with decreased cortical thickness in multiple brain areas and, conversely, that surgical or other invasive therapeutic interventions for CLBP lead to increased cortical thickness in the left DLPFC. Regarding functional changes, patients with CLBP were found to have abnormal DLPFC activation during cognitive challenge, and these cognitive task-related activation patterns normalized in a statistically significant manner following treatment. The authors conclude that CLBP alters brain structure and function in a reversible manner that can be measured using structural and functional imaging modalities. Their work provides additional evidence that the DLPFC is an important site of chronic pain physiology that, through functional imaging, may provide an objective means of measuring the experience of pain and its therapeutic modulation ( ).

In 2015 Čeko et al. demonstrated that, following therapeutic intervention for CLBP, partial recovery of abnormal insula and dorsolateral prefrontal connectivity to cognitive networks may be observed ( ). In this study the authors utilized rs-fMRI to examine the effects of CLBP on the connectivity of brain networks supporting cognitive function, and the modulatory effects of treatment on this network connectivity. The principal networks activated during performance of the cognitive task were identified in their task-fMRI data and used to define corresponding networks in the resting-state data. Comparison of the data between the pretreatment and posttreatment conditions led to identification of the bilateral insula as the region of abnormal cognitive connectivity in the resting state. A detailed connectivity analysis of the left DLPFC region was also undertaken, and revealed aberrant connectivity that was restored following treatment. In sum, the authors’ findings posit the bilateral insula and left DLPFC as important foci of abnormal cognition-related connectivity in CLBP ( ).

While clinical trials examining the efficacy of DLPFC targeting in neuromodulation for chronic pain are lacking, there is a growing number of studies reporting chronic pain alleviation following transcranial stimulation of the DLPFC. A 2009 case report by Arul-Anandam et al. describes one patient whose chronic cervical radicular pain was alleviated by transcranial direct-current stimulation to the DLPFC. More recently, Nardone et al. reported on the findings of a preliminary study looking at the effects of repetitive transcranial magnetic stimulation to the DLPFC on neuropathic pain in patients with spinal cord injury. The authors found a statistically significant improvement in neuropathic pain symptoms when compared to study participants who underwent sham treatment ( ). These findings suggest a possible role for direct stimulation of the DLPFC in the treatment of chronic pain syndromes.

Changes in hippocampal physiology have also been implicated as a biological substrate of chronic pain. The role of the hippocampus in the chronification of human pain was explored by Musto et al. in a 2014 study, using fMRI to examine hippocampal processing during a simple visual attention task. The authors compared fMRI signatures of intrinsic and extrinsic hippocampal connectivity in patients with subacute back pain and CLBP; both groups showed more extensive hippocampal connectivity than healthy controls. Upon examination of hippocampal connectivity in patients whose subacute back pain recovered as compared to those whose pain became chronic, the authors found distinct reorganization patterns in hippocampal–cortical connectivity. Specifically, when comparing patients with resolving subacute back pain to those whose pain tended to persist, they found significant differences over time in the connectivity of the hippocampus to the medial prefrontal cortex, the paracentral lobule, and the cingulate gyrus, suggesting that reorganization of hippocampal processing and hippocampal–cortical relationships may contribute to the transition from subacute to chronic pain ( ).

Since the 1970s the periventricular and periaqueductal gray matter (PVG/PAG) has become a target of interest in DBS surgery for the treatment of chronic pain, including intractable low back pain ( ). In a 2005 metaanalysis of DBS for pain relief, Bittar et al. found that targeting the PVG/PAG provided the highest rate of long-term pain alleviation, at 79% of the studied populations; this rate rose to 87% when the sensory thalamus/internal capsule were included in the PVG/PAG targeting ( ). Despite these clinical outcomes, the mechanism of analgesia in this area of neuromodulation remains incompletely understood. Only recently has the role of the PAG, its functional connectivity, and descending modulation in CLBP become another area of functional imaging research.

In 2014 Yu et al. performed a novel PAG seed-based functional connectivity (FC) analysis of fMRI data from patients with CLBP versus healthy controls. The authors sought to elucidate the connectivity maps in low versus high pain conditions versus healthy controls within a CLBP cohort at differing back pain intensities. In their methodological paradigm, participants underwent fMRI studies of the brain both prior to and following a series of clinical maneuvers aimed at exacerbating the pain experience. An FC analysis was then performed using the right ventrolateral PAG as the FC seed; the authors rationalized this choice of location on the basis of prior findings that this region demonstrates significant fMRI signal increase in other pain paradigms, that it is located in a region thought to play an important role in opioid antinociception, and that the PAG had previously demonstrated FC to the ventral medial prefrontal cortex and rostral anterior cingulate cortex. The authors found that FC between the PAG and the ventral medial prefrontal cortex/rostral anterior cingulate cortex was elevated in patients with CLBP as compared to healthy controls, but in the CLBP group there were negative correlations between pain ratings and this FC after performance of the pain-exacerbating maneuver. Moreover, duration of CLBP in this group was negatively correlated with FC between the PAG and both the insula and the amygdala before the pain-inducing maneuver. The investigators concluded that their findings were in line with the impairments of the descending pain modulation reported in patients with CLBP; specifically, CLBP was found to elicit abnormal FC in the PAG-centered pain modulation network during rest ( ). However, PVG/PAG DBS largely disappeared after disappointing results in two multicenter trials ( ).

The significance of this article by was revisited in a review by . The authors place the study in the context of the uniqueness of its methodology, clinical exacerbation of pain, for finding altered PAG connectivity, as well as from the perspective of the challenges inherent in studying this region and the impact of the findings on future pain research. The authors provide a summary diagram contextualizing the findings of Yu et al. within a perspective of other work on PAG FC, and conclude that the direction of PAG FC change in the study by Yu et al. during the “exacerbated” condition is in agreement with similar findings in other literature, but that during the “nonexacerbated” state (black arrows) the direction of PAG FC appears to oppose the direction of change in acute pain in both the study in question and other literature ( Fig. 46.3 ). In an effort to explain this potentially perplexing latter finding, Hemington and Coulombe suggest decreased FC during increased pain intensity is expected during the acute pain experience, whereas chronification of pain may involve emotional factors that have distinct underlying mechanisms; their suggestion underscores the concept that a single brain region can be involved in a multitude of functions, and alterations in its connectivity can be multicausal ( .

A brief summary of periaqueductal gray (PAG, large oval ) functional connectivity (FC) in relation to acute or chronic low back pain. Gray arrows indicate findings after a noxious stimulus that triggers acute back pain. Black arrows indicate FC during chronic back pain. vmPFC , ventro-medial prefrontal cortex.

Adapted from Hemington, K.S., Coulombe, M.A., October 2015. The periaqueductal gray and descending pain modulation: why should we study them and what role do they play in chronic pain? J. Neurophysiol. 114 (4), 2080–2083.

The authors underscore that regions of the brainstem such as the PAG are an oft-neglected area in fMRI research given the challenges of studying these regions, e.g., potential confounding effects of neighboring regions of cerebrospinal fluid or blood flow, but they assert that understanding the role of descending systems is a critical aspect of understanding the larger context of pain experience.

Trigeminal Neuralgia

FMRI has theoretical uses for peripheral pain syndromes. Specifically, several studies have assessed the functional connectome in patients with trigeminal neuralgia. In a prospective study of 15 patients with trigeminal neuralgia of the V2 or V3 distributions, stimulation of a trigger zone that resulted in an allodynic response had an fMRI signature that was significantly distinct from a patient whose typical trigeminal pain was not elicited by trigger zone stimulation ( ) ( Fig. 46.4 ). Evoked pain was associated with bilateral activation of the primary and secondary somatosensory cortices, prefrontal cortex, and anterior cingulate cortex, contralateral activation of the anterior insula and thalamus, and ipsilateral activation of the medium cingulate cortex and caudal medulla in the spinal trigeminal nucleus, as well as activation of the PAG. There was also significant activation of the hippocampal/parahippocampal regions, precentral cortex, supplementary motor area, putamen, and cerebellum. When the patients were stimulated on the side of the face contralateral to the trigger zone, no such allodynic responses were noted clinically and fMRI activation was found in the bilateral somatosensory cortices and the precentral cortex, as well as the contralateral supplementary motor area. Furthermore, in patients who underwent surgical treatment of their trigeminal neuralgia, postoperative fMRI following retrogasserian radiofrequency ablation highlighted activation profiles similar to the pretreatment nonaffected side; that is, activation of only the bilateral somatosensory cortices, precentral cortices, and contralateral supplementary motor area.

Sequential fMRI images demonstrating brain areas that are activated (in yellow) during painful (panel A) or nonpainful stimuli (panel B). ACC , anterior cingulate cortex; ant. insula , anterior insular cortex; MCC , medial cingulate cortex; post. insula , posterior insular cortex; S1 , primary somatosensory cortex; S2 , secondary somatosensory cortex; SMA , supplementary motor area; SpV , spinal trigeminal nucleus; thal , thalamus.

Adapted from Moisset, X., Villain, N., Ducreux, D., Serrie, A., Cunin, G., Valade, D., Calvino, B., Bouhassira, D., February 2011. Functional brain imaging of trigeminal neuralgia. Eur. J. Pain 15 (2), 124–131, used with permission.

Activation of the spinal trigeminal nuclei has long been identified in fMRI in patients with active trigeminal neuralgic attacks, in both animal models and human studies ( ) ( Fig. 46.4 ). This is proposed to relate to hyperexcitability or sensitization of the nociceptive trigeminal neurons. In trigeminal neuralgia such a pathologic response has been hypothesized to be due to focal irritation and demyelination of the trigeminal roots by adjacent vascular loops, which leads to the generation of ectopic neural discharges and cross-activation between large and small afferent fibers in the Gasserian ganglion ( ). This response can be visualized on fMRI via activation of the spinal trigeminal nuclei.

Further functional imaging studies have been employed to assess changes in patients with trigeminal neuralgia. One study by Obermann et al. in 2013 utilized voxel-based morphometry to assess structural differences between 60 patients with trigeminal neuralgia and 49 healthy controls. The group identified grey matter volume reduction in patients with trigeminal neuralgia, specifically in the somatosensory cortices, thalamus, insula, anterior cingulate cortex, cerebellum, and dorsolateral prefrontal cortex. Furthermore, the degree of volume loss was correlated with duration of symptoms.

A 2013 study by Liu et al., assessed the use of DTI in patients with trigeminal neuralgia. They found significantly decreased FA and increased radial diffusivity on the affected trigeminal root entry zone when compared to the unaffected side ( Fig. 46.5 ). These parameters have been used to differentiate demyelinating changes as compared to axonal injury, thus these results suggest that trigeminal neuralgic pain is generated by demyelination of the trigeminal nerve root entry zone rather than by axonal loss. The authors further suggest that FA can be used as a diagnostic biomarker for patients undergoing workup for trigeminal neuralgia.

A representative color fractional anisotrophy image demonstrating the regions of interest used for quantitative analysis of the root entry zone of the trigeminal nerve.

Adapted from Liu, Y., Li, J., Butzkueven, H., Duan, Y., Zhang, M., Shu, N., Li, Y., Zhang, Y., Li, K., May 2013. Microstructural abnormalities in the trigeminal nerves of patients with trigeminal neuralgia revealed by multiple diffusion metrics. Eur. J. Radiol. 82 (5), 783–786, used with permission.

MR tractography has also been used to assess posttreatment changes in patients who have undergone focal radiosurgery for trigeminal neuralgia ( ). The authors demonstrated a significant decrease in FA and a significant increase in radial diffusivity following targeted radiosurgery. These results suggest the underlying effect of radiation on myelin. Furthermore, the authors correlated a recovery of FA to pretreatment baselines with symptomatic recurrence of trigmeninal neuralgia.

Motor cortex stimulation (MCS) has also been utilized as a potential therapeutic option for intractable facial pain syndromes. Esfahani et al. described a unique synthesis of fMRI and MCS for the treatment of trigeminal neuralgia after rhizotomy secondary to tumor and postherpetic neuralgia ( ). In this study the authors utilized fMRI to localize the facial areas on the motor cortex. This allowed for exquisite target selection with MCS electrodes. Although the study utilized a small cohort of patients (n = 3), it demonstrated fMRI’s utility in enabling neurosurgeons to be confident with treatment. MCS for phantom limb pain utilizing fMRI to detect shoulder and stump cortical activated areas has also been described ( ).

Functional Magnetic Resonance Imaging

The advent of fMRI in the 1990s allowed clinicians to examine changes in brain activity after a stereotyped task ( ). The first imaging methods used intravenous gadolinium contrast or the BOLD technique, which detects levels of deoxyhemoglobin in stimulated brain ( ). When subjected to a simple mechanical or visual task, areas of eloquent cortex become activated and subsequently utilize oxygen to generate energy. As a result, local levels of deoxyhemoglobin increase in concentration, and these depots of deoxyhemoglobin have inherent paramagnetic properties that can be detected with BOLD imaging ( ). Since then there have been significant advancements in the imaging capabilities of fMRI, including the ability to examine brain activity at rest with rs-fMRI ( ). When compared to BOLD imaging, which depicts brain activity during stereotyped tasks, rs-fMRI demonstrates brain activity without an applied stimulus, or at rest. Although intraoperative, direct electrical stimulation remains the gold standard for mapping functional cortex, these new fMRI technologies allow better preoperative planning and intraoperative guidance.

Fomberstein et al. described several anatomic regions in the brain that have been implicated in pain perception ( ). These areas, termed the “pain matrix,” include the primary and secondary somatosensory cortex, anterior cingulate cortex, and insula ( ). Abnormal functional connectivity between these areas has been shown to be highly correlated with several chronic pain syndromes ( ). The authors found that functional connectivity between the prefrontal cortex, insula, and anterior cingulate cortex could be used as a marker for chronic pain that correlates highly with poor treatment response to opioid therapy ( ) ( Fig. 46.1 ) . The authors noted several interesting patterns of activation, displayed on fMRI in several chronic pain states. For example, patients with chronic low back pain (CLBP) had increased activation of the medial prefrontal and anterior cingulate cortex when compared to normal controls. After 2 weeks of treatment, connectivity of the dorsal median prefrontal cortex and insula was seen in the treatment-refractory patient group ( ). In addition, higher connectivity between the dorsal median prefrontal cortex and insula was directly correlated with decreased responsiveness to placebo treatment in this study ( ). These results support the long-held belief that chronic pain syndromes are a multifaceted entity involving more than just noxious stimulus of peripheral or autonomic nerves ( ).

Brain regions involved in (A) pressure pain processes and (B) pressure pain > nonpainful pressure as measured by BOLD fMRI from simultaneous PET/MRI scans. BOLD , blood oxygenation level dependent; PET , positron emission tomography; MRI , magnetic resonance imaging.

Adapted from Wey, H.Y., Catana, C., Hooker, J.M., et al., November 15, 2014. Simultaneous fMRI-PET of the opioidergic pain system in human brain. Neuroimage 102 (Pt 2), 275–282, with permission.

As most clinicians are well aware, patients with complex chronic pain syndromes can be quite challenging to manage and often require multidisciplinary care. These results suggest that fMRI techniques could be used to predict treatment response to analgesia, thus creating a framework to build patient-specific analgesic regimens that are supported by connectivity. Connectivity could be used to predict opioid response and probability of treatment success/failure, set thresholds for treatment duration, and monitor treatment progression. Unlike conventional MRI, fMRI can be used to obtain objective information about subjective clinical data, such as pain. With this in mind, there have been more recent efforts to tailor fMRI sequences to detecting areas that become active after a noxious stimulus, in hope that this data can be used to “diagnose” pain.

In 2013 Wager et al. were the first to develop an fMRI-based neurologic signature that could detect physiological pain. In response to a noxious stimulus, Wager’s pain signature demonstrated increased activity in the primary and secondary somatosensory cortices that was over 90% sensitive and specific for the experimental noxious stimulus ( ). The study investigators hypothesized that the signature could be applied to clinical situations when patients are unable to communicate pain or clinicians suspect incorrect self-reporting of pain.

In another study conducted by Loggia et al. in 2014, fMRI was used to characterize pain physiology and circuit pathways in 13 patients with fibromyalgia. The investigators found that these patients repeatedly experienced an exaggerated response to a painful stimulus when compared to normal volunteers. In addition, fibromyalgia patients demonstrated reduced reward/punishment signaling, thought to be due to decreased activity in the periaqueductal gray and ventral segmental areas of the midbrain. They hypothesized that suppression of activity in these regions may explain why fibromyalgia patients do not respond well to traditional pain therapies ( ). In fact, the treatment of fibromyalgia often requires a multidisciplinary clinical approach and the judicious use of neuroleptic, antidepressant, and occasionally antiepileptic mediations ( ). The results of these studies emphasize the difficulty in treating chronic pain disorders with opioid medications alone, and may suggest that a multimodality approach is needed.

Spinal Cord Stimulation

Rs-fMRI has also been examined in the context of characterizing pain relief from spinal cord stimulation (SCS). Deogaonkar et al. utilized fMRI in 10 patients who underwent thoracic SCS implantation for complex regional pain syndrome or neuropathic leg pain ( ). Patients underwent imaging in both the on- and off-stimulation states. The investigators found that pain relief from SCS reduced resting-state connectivity in the left frontal insula, right primary/secondary somatosensory cortex, and default mode network (DMN) ( ). The investigators further hypothesized that SCS exerted its analgesic effects through promoting decreased connectivity between the somatosensory and limbic areas of the brain. These results are intriguing, because they suggest that the effects of SCS are more complex than simply blocking the propagation of pain stimuli from peripheral nerves. In fact, the effects of SCS trigger changes in the severity and the brain’s processing of pain at the level of brain circuitry.

Although these studies suggest that fMRI techniques may be utilized to characterize pain objectively, clinical investigators underscore the need for further studies with larger populations before determining the clinical utility of these sequences for pain syndromes. Functional imaging techniques have been used for identification of eloquent areas in tumor and epilepsy neurosurgery, but the application of these techniques in neuromodulation is still under investigation ( ). Certainly, as more advanced and pain-specific fMRI techniques are developed, functional connectivity patterns between various pain-generating areas of the brain could be better detected and treated ( ).

Arterial Spin Labeling

Traditionally used in the imaging of brain tumors, the ASL MRI sequence characterizes changes in cerebral blood flow (CBF). In 2011 Wasan et al. demonstrated measurable increases in regional CBF to the somatosensory, prefrontal, and insular cortices in patients with CLBP subjected to a straight-leg-raising test. The authors also discovered that the same straight-leg maneuver did not create detectible ASL changes in healthy age-matched controls. Patients with CLBP also demonstrated decreased connectivity between the medial prefrontal cortex and the DMN ( ). The authors hypothesized that higher levels of connectivity between the medial prefrontal cortex and the DMN in the control population may enhance pain-coping mechanisms.

In a study exploring the role for pulsed-continuous ASL (PC-ASL) in a thermal stimulation paradigm, Maleki et al. compared BOLD techniques to PC-ASL in 12 healthy subjects ( ). Using PC-ASL, the authors found specific activation patterns in response to a thermal painful stimulus. The insula, thalamus, hippococampus, amygdala, anterior cingulate, primary and secondary somatosensory cortex, precentral gyrus, and precuneus all had recognizable activation patterns ( ). The authors conclude that PC-ASL may be used to study acute pain, and hypothesize that this imaging modality might provide a more suitable imaging approach to measure brain activity in patient populations where blood flow, volume, or oxygen extraction are altered ( ).

More recently, ASL imaging has been utilized in an effort to identify pain-specific regions of cortical activation and apply ASL to clinical scenarios. recognized the dorsal posterior insula as a fundamental structure for human pain processing, and posited that it may be a potential therapeutic target. Using controlled, slowly varying nociceptive input with a capsaicin-induced thermal stimulation model in 17 awake, healthy human subjects, the authors quantified cortical activation using an optimized ASL method. They subsequently evaluated the correlation between absolute CBF and pain scores during the experimental time course, and determined that, from a whole-brain analysis, only the contralateral dorsal posterior insula showed statistically significant changes ( ). The authors further clarify the putative role of this brain region for tracking pain intensity, and contribute to an evolving discussion about the role of ASL in elucidating the precise circuitry of pain ( ).

Combining ASL data for detection of pain-related changes in CBF with multivariate machine-learning techniques for analysis of prognostic and diagnostic markers of pain, O’Muircheartaigh et al. demonstrated the feasibility of time-efficient probabilistic prediction of clinically relevant pain ( ). The authors assessed preoperative and postoperative ASL data and pain scores in 20 male volunteers requiring bilateral lower-jaw third molar extraction. The data from all sessions was fed into an independent Gaussian process binary classifier that successfully discriminated postsurgical from presurgical states with nearly 95% accuracy. The authors concluded that multivariate pattern classification methods can be applied to ASL data to provide clinically relevant predictions regarding moderate to severe postoperative pain. On the basis of these findings, they further suggest a role for functional neuroimaging as an objective complement to subjective pain measurements in a perioperative patient population ( ).

Magnetic Resonance Tractography

Numerous investigators have observed recurrent patterns of anatomic and physiologic changes in white-matter circuitry of patients with chronic pain ( ). Magnetic resonance diffusion-weighted imaging (dMRI) has been used to explain and further characterize these differential structural patterns. Furthermore, dMRI can be used to estimate the trajectories of white-matter fiber pathways, a technique known as fiber tractography. Experimentally, dMRI has been shown to be a sensitive diagnostic imaging tool to diagnose a wide variety of chronic pain disorders, ranging from lumbar radiculopathy to migraine headaches ( ).

Tractography assesses the restricted random diffusion of water molecules in axons to characterize their microstructure. It is generally composed of the aggregate information of two different imaging sequences, diffusion tensor imaging (DTI) and fractional anisotropy (FA). DTI detects the movement of water through axons and FA characterizes their direction. The structural connectivity of the brain can then be reconstructed and visualized with dMRI. Abnormal patterns of white-matter circuitry, implicated in the development of chronic pain disorders, can then be diagnosed and treated. For example, Kovanlikaya et al. found that pain fibers traversing the nucleus ventrocaudalis of the thalamus could be visualized with dMRI and used as a target for deep brain stimulation (DBS) surgery ( ). Mansour et al. observed differences in the patterns of structural connectivity in the prefrontal cortex of patients who developed CLBP after an inciting event. They hypothesized that these structural differences, characterized with dMRI ( ), which existed prior to the noxious stimulus could predict the patient’s predisposition to chronic pain.

In addition to examining spinothalamocortical pain pathways, dMRI has been shown to be a sensitive indicator of spinal cord and nerve root compression ( Fig. 46.2 ) ( ). Shi et al. found that the mean FA values in compressed nerve roots were significantly decreased when compared to normal uncompressed nerve roots. These findings were further supported by Oikawa et al., who hypothesized that dMRI may offer quantitative values for the severity of nerve root compression, which can be used as a diagnostic tool for lumbar radiculopathy ( ). Similar observations were used by Fujiwara et al. to image and diagnose trigeminal neuralgia from neurovascular compression. In this study, patients with trigeminal neuralgia from suspected neurovascular compression demonstrated lower mean FA values compared to healthy controls ( ).

Fusion image made up of the diffusion tensor tractography of the lumbar nerve (L4, L5, S1) and T2-weighted imaging. Interruption in the right L5 nerve at the foramen ( white arrow ).

Adapted from Eguchi, Y., Ohtori, S., Suzuki, M., et al., February 2016. Diagnosis of lumbar foraminal stenosis using diffusion tensor imaging. Asian Spine J. 10 (1), 164–169, with permission.

This data provides clinicians with an objective, quantitative tool to measure the degree of nerve root compression throughout the lumbar spine. In disease states such as degenerative lumbar spondylosis, where there are numerous and varying severities of nerve root compression, FA values could potentially be used in tandem with clinical exam and history to guide candidate selection for surgery. In addition, FA values can be utilized to quantify and assess the degree of decompression after surgery and potentially be used as an outcome marker for prospective outcome studies.

Chronic Low Back Pain

Over the past decade the structural and functional imaging signatures of brain physiology in patients with CLBP and its modulation and treatment have been a growing area of investigation. CLBP and other types of chronic pain have been associated with both anatomical and functional changes in the cerebral cortex. Several imaging and neurophysiological studies have implicated the dorsolateral prefrontal cortex (DLPFC) as one common site of such changes. For example, using an fMRI-based protocol, Seminowicz and Davis demonstrated in 2006 that pain catastrophization, i.e., the maladaptive exaggeration of a painful experience, was not related to activity in somatosensory cortices of the parietal lobe, but rather to cortical regions such as the dorsolateral prefrontal and insular cortices that subserve affective, attention-related, and motor aspects of pain ( ). Their findings were posited to support the framework of an attention model of pain catastrophization where, despite engagement of a cortical vigilance network during mild pain, diminished prefrontal cortical modulation inhibits pain suppression during more intense painful stimuli. Furthermore, the authors speculate that catastrophization may underlie the mechanistic progression from acute pain to chronic pain.

In 2010 Fierro et al. reported on an exploration of the effect of capsacin-induced pain and the modulatory influences of left DLPFC transcortical magnetic stimulation on motor corticospinal and intracortical excitability; the authors conclude that tonic pain exerts a modulatory influence of motor intracortical excitability and that left DLPFC stimulation can have analgesic effects on otherwise painful stimuli. In the same year Krummenacher et al. reported on the ability of DLPFC stimulation to interrupt placebo analgesia in a heat-pain paradigm ( ).

There is also mounting evidence that some people with chronic pain, such as longstanding back pain, exhibit abnormal cortical function and cognitive impairment. Findings from a 2007 study by Seminowicz and Davis suggest that the demands placed on pain-associated cognitive networks may underlie this cognitive impairment ( ). In 2011 Seminowicz et al. reported on the findings of a prospective study assessing the reversibility of neuroanatomical and functional abnormalities in patients with CLBP, and the dependence of these changes on treatment outcomes ( ). In their longitudinal study, the authors utilized 3 T MRI for structural and functional brain imaging in patients with CLBP. In the experimental group undergoing intervention (n = 14), the imaging studies were performed prior to and 6 months after either spine surgery or interventional pain procedures. The experimental cases were compared to 10 healthy controls. Cortical thickness analysis was performed using structural MRI studies, and participants also underwent fMRI scans during the performance of a cognitively demanding task. The authors found that CLBP was associated with decreased cortical thickness in multiple brain areas and, conversely, that surgical or other invasive therapeutic interventions for CLBP lead to increased cortical thickness in the left DLPFC. Regarding functional changes, patients with CLBP were found to have abnormal DLPFC activation during cognitive challenge, and these cognitive task-related activation patterns normalized in a statistically significant manner following treatment. The authors conclude that CLBP alters brain structure and function in a reversible manner that can be measured using structural and functional imaging modalities. Their work provides additional evidence that the DLPFC is an important site of chronic pain physiology that, through functional imaging, may provide an objective means of measuring the experience of pain and its therapeutic modulation ( ).

In 2015 Čeko et al. demonstrated that, following therapeutic intervention for CLBP, partial recovery of abnormal insula and dorsolateral prefrontal connectivity to cognitive networks may be observed ( ). In this study the authors utilized rs-fMRI to examine the effects of CLBP on the connectivity of brain networks supporting cognitive function, and the modulatory effects of treatment on this network connectivity. The principal networks activated during performance of the cognitive task were identified in their task-fMRI data and used to define corresponding networks in the resting-state data. Comparison of the data between the pretreatment and posttreatment conditions led to identification of the bilateral insula as the region of abnormal cognitive connectivity in the resting state. A detailed connectivity analysis of the left DLPFC region was also undertaken, and revealed aberrant connectivity that was restored following treatment. In sum, the authors’ findings posit the bilateral insula and left DLPFC as important foci of abnormal cognition-related connectivity in CLBP ( ).

While clinical trials examining the efficacy of DLPFC targeting in neuromodulation for chronic pain are lacking, there is a growing number of studies reporting chronic pain alleviation following transcranial stimulation of the DLPFC. A 2009 case report by Arul-Anandam et al. describes one patient whose chronic cervical radicular pain was alleviated by transcranial direct-current stimulation to the DLPFC. More recently, Nardone et al. reported on the findings of a preliminary study looking at the effects of repetitive transcranial magnetic stimulation to the DLPFC on neuropathic pain in patients with spinal cord injury. The authors found a statistically significant improvement in neuropathic pain symptoms when compared to study participants who underwent sham treatment ( ). These findings suggest a possible role for direct stimulation of the DLPFC in the treatment of chronic pain syndromes.

Changes in hippocampal physiology have also been implicated as a biological substrate of chronic pain. The role of the hippocampus in the chronification of human pain was explored by Musto et al. in a 2014 study, using fMRI to examine hippocampal processing during a simple visual attention task. The authors compared fMRI signatures of intrinsic and extrinsic hippocampal connectivity in patients with subacute back pain and CLBP; both groups showed more extensive hippocampal connectivity than healthy controls. Upon examination of hippocampal connectivity in patients whose subacute back pain recovered as compared to those whose pain became chronic, the authors found distinct reorganization patterns in hippocampal–cortical connectivity. Specifically, when comparing patients with resolving subacute back pain to those whose pain tended to persist, they found significant differences over time in the connectivity of the hippocampus to the medial prefrontal cortex, the paracentral lobule, and the cingulate gyrus, suggesting that reorganization of hippocampal processing and hippocampal–cortical relationships may contribute to the transition from subacute to chronic pain ( ).

Since the 1970s the periventricular and periaqueductal gray matter (PVG/PAG) has become a target of interest in DBS surgery for the treatment of chronic pain, including intractable low back pain ( ). In a 2005 metaanalysis of DBS for pain relief, Bittar et al. found that targeting the PVG/PAG provided the highest rate of long-term pain alleviation, at 79% of the studied populations; this rate rose to 87% when the sensory thalamus/internal capsule were included in the PVG/PAG targeting ( ). Despite these clinical outcomes, the mechanism of analgesia in this area of neuromodulation remains incompletely understood. Only recently has the role of the PAG, its functional connectivity, and descending modulation in CLBP become another area of functional imaging research.

In 2014 Yu et al. performed a novel PAG seed-based functional connectivity (FC) analysis of fMRI data from patients with CLBP versus healthy controls. The authors sought to elucidate the connectivity maps in low versus high pain conditions versus healthy controls within a CLBP cohort at differing back pain intensities. In their methodological paradigm, participants underwent fMRI studies of the brain both prior to and following a series of clinical maneuvers aimed at exacerbating the pain experience. An FC analysis was then performed using the right ventrolateral PAG as the FC seed; the authors rationalized this choice of location on the basis of prior findings that this region demonstrates significant fMRI signal increase in other pain paradigms, that it is located in a region thought to play an important role in opioid antinociception, and that the PAG had previously demonstrated FC to the ventral medial prefrontal cortex and rostral anterior cingulate cortex. The authors found that FC between the PAG and the ventral medial prefrontal cortex/rostral anterior cingulate cortex was elevated in patients with CLBP as compared to healthy controls, but in the CLBP group there were negative correlations between pain ratings and this FC after performance of the pain-exacerbating maneuver. Moreover, duration of CLBP in this group was negatively correlated with FC between the PAG and both the insula and the amygdala before the pain-inducing maneuver. The investigators concluded that their findings were in line with the impairments of the descending pain modulation reported in patients with CLBP; specifically, CLBP was found to elicit abnormal FC in the PAG-centered pain modulation network during rest ( ). However, PVG/PAG DBS largely disappeared after disappointing results in two multicenter trials ( ).

The significance of this article by was revisited in a review by . The authors place the study in the context of the uniqueness of its methodology, clinical exacerbation of pain, for finding altered PAG connectivity, as well as from the perspective of the challenges inherent in studying this region and the impact of the findings on future pain research. The authors provide a summary diagram contextualizing the findings of Yu et al. within a perspective of other work on PAG FC, and conclude that the direction of PAG FC change in the study by Yu et al. during the “exacerbated” condition is in agreement with similar findings in other literature, but that during the “nonexacerbated” state (black arrows) the direction of PAG FC appears to oppose the direction of change in acute pain in both the study in question and other literature ( Fig. 46.3 ). In an effort to explain this potentially perplexing latter finding, Hemington and Coulombe suggest decreased FC during increased pain intensity is expected during the acute pain experience, whereas chronification of pain may involve emotional factors that have distinct underlying mechanisms; their suggestion underscores the concept that a single brain region can be involved in a multitude of functions, and alterations in its connectivity can be multicausal ( .

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

Fundamentals of Burst Stimulation of the Spinal Cord and Brain Advances in Invasive Brain–Computer Interface Technology and Decoding Methods for Restoring Movement and Future Applications Spinal Cord Stimulation for the Treatment of Low Back Pain Epilepsy : Anatomy, Physiology, Pathophysiology, and Disorders Colonic Electrical Stimulation for Constipation Transcranial Alternating Current Stimulation and Transcranial Random Noise Stimulation

Stay updated, free articles. Join our Telegram channel

Join